When determining the trajectory of the dummy head during a vehicle crash test, the head is not visible for all cameras during the whole movement, since it, e.g., dips into the airbag. Another possibility is to calculate the trajectory with acceleration and gyro sensor data. When using low-cost inertial measurement units, the calculated trajectory differs from the video analysis by 0.15 metres. The idea is to model possible electrical sensor errors and inaccurate known initial conditions and determine them in an optimisation process in which the video trajectory is taken as the target function. This paper deals with the optimisation process to determine if the optimisation function has a unique minimum. Based on that, several optimisation algorithms are compared and three of them are selected for a detailed comparison. The best of them is selected to show the calculation of a trajectory for a real world crash test.
24
E.C. Klein et al.Keywords: dummy trajectory calculation; sensor parameter optimisation; optimisation function classification; optimisation algorithm comparison.Reference to this paper should be made as follows: Klein, E.C., Sinz, W., Moser, J., Greimel, R., Raguse, K., von Middendorff, C. and Steiner, C. (2016
Comparison of optimisation strategies
25Christina Steiner is currently finishing her PhD thesis, which deals with the development and application of an endoscope measurement system for high speed filming in narrow spaces. She received her MSc in 2009 from the University of Karlsruhe with her thesis about the implementation of a 3D video analysis system in the vehicle safety department at the AUDI AG. Her focus is on optical measurement technology and she is leading the team for the development of test-and measurement techniques.